1. Microglia Ontogeny, Distribution and Turnover

Microglia—the innate immune cells of the central nervous system (CNS)—belong to CNS myeloid cells that encompass microglia and CNS border-associated macrophages (BAMs). Early studies using immunohistochemistry demonstrated that microglia represent about 10% of the adult brain cell population and are present in all main CNS structures, although these cells are not uniformly distributed [1]. In initial studies, staining for microglia markers F4/80 and receptors of FcIgG1/2b (Fc, fragment crystallizable), as well as type-three complement (Mac-1) showed two types of positive cells in the adult mouse brain: microglia and cells associated with the choroid plexus, ventricles, and leptomeninges [2]. There is regional diversity in the abundance of microglia; more positive cells are found in the gray matter than the white matter. Microglia are numerous in the hippocampus, olfactory telencephalon, basal ganglia, and substantia nigra. The less densely populated areas include fiber tracts, cerebellum, and the brainstem, while the cerebral cortex, thalamus, and hypothalamus have average cell densities. The proportion of microglia varies from 5% in the cortex and corpus callosum to 12% in the substantia nigra.

Regional heterogeneity of microglia has been detected in the adult human brain, with lower densities in the gray matter compared with the white matter, and substantially lower densities in the cerebellar cortex compared to the substantia nigra ^[3][4][5][3,4,5]. Common immunological markers include: CD (cluster of differentiation) 45, CD68, HLA-DR (a MHC class II cell surface receptor), IBA-1 (Ionized calcium-binding adaptor molecule 1). Those microglia slowly renew at a median rate of 28% per year, and only 2% of microglia are thought to be proliferating at a given time ^[6][7][6,7]. Most studies in humans assessed microglial populations in embryonic/fetal tissues with immunological (CD45, MHCII, CD68, IBA-1) or histochemical markers (RCA-1) that do not distinguish between microglia and non-parenchymal macrophages.

The issue of microglia maintenance in adulthood has been addressed in studies using bone marrow (BM) cell transplantation and parabiosis. The BM transplantation experiments in rats showed that solely perivascular macrophages are replaced and not the ramified microglia in the brain parenchyma [8]. In female human patients who underwent sex mismatched BM transplantation, only donor-derived perivascular macrophages were detected [9]. In the early cell transfer experiments with transplantation of the BM hematopoietic cells expressing green fluorescent protein (GFP), some GFP-expressing parenchymal microglia were found in the cerebellum, striatum, and hippocampus, suggesting a partial substitution by peripheral cells ^[10][11][10,11]. However, in later studies, in which an animal head was shielded to protect from irradiation, any significant infiltration of BM-derived cells into the brain was not observed, suggesting that the spotted substitution was due to radiation [12]. The experiments on parabiotic mice, in which the turnover of hematopoietic cells for prolonged periods can be studied, demonstrated a lack of microglia progenitor recruitment from the circulation in denervation or CNS neurodegenerative disease. Further studies combining parabiosis and myeloablation showed that recruited monocytes do not persist in the CNS [13]. The expression of a progenitor marker, nestin, is corroborating evidence that microglial progenitors were not hematopoietic cells. All those findings pointed to the persistence of microglia in CNS and a lack of the significant contribution from BM-derived monocytes [14].

In vivo lineage tracing studies, using the fractalkine receptor encoding gene Cx3cr1gfp/+ knock-in mice, established that microglia have a different ontogeny from mononuclear phagocytes and colonize CNS early in the development. Several studies demonstrated that microglia are derived from primitive hematopoietic progenitors (c-Kit^loCD41^lo progenitors) that originate around embryonic day 7.25 (E7.25) in the yolk sac ^[15][16][17][15,16,17]. A recent study found two populations: non-Hoxb8 microglia and Hoxb8 microglia, with the latter derived from the second wave of yolk sack hematopoiesis infiltrating the murine brain around E12.5 [18]. In the zebrafish embryo, microglia derive from c-myb-independent erythro-myeloid progenitors but are replaced by c-myb-dependent hematopoietic stem cells (HSC) after birth [19]. These two studies suggest a second wave of proliferation and CNS colonization by microglial progenitors in the early CNS development. All the data point to the origin of microglia from the yolk sack hematopoiesis, early CNS colonization, and the lack of significant input from HSCs in adulthood. Similarly to microglia, BAMs are derived from hematopoietic precursors during embryonic development and establish stable populations, with the exception of choroid plexus macrophages, which have a shorter life span and are replenished from blood-borne monocytes [20].

Due to different ontogeny and location, microglia acquire a different gene expression signature than BAMs and peripheral macrophages [21]. In adulthood, microglia are dependent on constant stimulation of colony-stimulating factor-1 (Csf1) receptors. In both mice and humans, interleukin-34 (IL-34), an alternative ligand for Csf-1 receptor produced by neurons in the brain, is essential for microglia maintenance ^[22][23][22,23].

The initial colonization of the brain by microglia corresponds to development of its vascularization, although microglia may enter the brain via brain ventricles or across meninges [24]. In the human brain, microglia colonization coincides with the vascularization, radial glia formation, neuronal migration, and myelination [25]. The density of microglia in the developing CNS is two times higher than in the adult CNS, and the time-regulated decline in microglia number involves increased apoptosis and reduced proliferation [26]. Microglia acquire a definitive local density soon after birth, after a wave of microglial proliferation within the first 2 postnatal weeks followed by a decline by 50% between the third and sixth postnatal weeks [26].

Studies of microglia in human brain sections from embryos and fetuses from early gestational weeks (gw) showed that microglial cells penetrate into and spread throughout the cortical gray and white matter with the specific spatiotemporal pattern during the first 2 trimesters of gestation. Amoeboid microglia positive for IBA1 (ionized calcium-binding adapter molecule 1), CD68, and CD45 were present in the forebrain starting from 4.5 gw. They penetrate the telencephalon and diencephalon via the meninges, choroid plexus, and ventricular zone. A second wave of microglial cells penetrates the brain via the vascular route at about 12-13 gw and remains confined to the white matter ^[27][28][27,28].

There are some indications that microglia turnover in mice is slowed down with aging. Studies on C57BL/6J-Iba1-eGFP mice (a gene encoding enhanced green fluorescence protein under the iba1 promoter) using in vivo 2-photon microscopy showed that microglial cells in the neocortex exhibit age-related changes: increases of a soma volume, shortening of microglial processes, reduction of motility of processes, and changes in tissue distribution [29]. Another mouse strain CD11b-CreERT2;R26-tdTomato expresses the red fluorophore tdTomato under the control of the cd11b gene promoter. In vivo single-cell imaging in triple-transgenic CD11b-CreERT2;R26-tdTomato;APPPS1 mice (an Alzheimer disease model, double transgenic mice expressing a chimeric mouse/human amyloid precursor protein (Mo/HuAPP695swe) and a mutant human presenilin 1 (PS1-dE9) showed that ≈20% microglia disappear in areas without amyloid deposits over the 6-month imaging period, whereas the microglia loss in the wild-type mice was ~13% over the same imaging period. The results suggested that the increase in microglia around amyloid deposits results from microglial proliferation in plaque-free areas and migration toward the plaques. The newly appearing microglia were derived from the division of resident microglial cells [30].

2. Microglia Heterogeneity and Function in Health and Disease

Microglia play a pivotal role in brain development, immune defense, and the maintenance of CNS homeostasis [31]. Microglia, with ramified and motile processes, surveil the brain parenchyma for dysfunction, infection, or damage [32]. Microglia undergo morphological changes under pathological conditions that can be categorized and quantified for parameters such as soma size, cell ramification, branching complexity, and shape. Generation of transgenic Cx3cr1^+/GFP mice with a one allele coding for the CX3C chemokine receptor (Cxc3cr1) replaced with GFP (green fluorescence protein) allowed to study morphological changes of microglia. For example, severe ischemia in brain slices from Cx3cr1^+/GFP transgenic mice leads to pronounced de-ramification and the appearance of amoeboid-shaped cells [33].

The detection of pathogen-derived signals initiates microglial responses that on one hand instigate inflammation, but on the other hand attempt to resolve the injury, protect the CNS from the consequences of inflammation, and support tissue repair and remodeling ^[31][34][35][31,34,35]. The oversimplified classification, which divides microglia into the M1 inflammatory and the M2 pro-regenerative macrophages, fails to explain the diversity of myeloid subpopulations in the diseased brain. There were attempts to attribute both functions to microglia and the prevalence of each functional subpopulation to severity of the brain damage [36]. However, in diseases that affect the integrity of the blood–brain barrier, there are considerable increases in a number of macrophages due to influx of peripheral immune cells. The pressing question in the field is whether BM-derived monocytes accumulate in the brain and how they function under pathological conditions.

Immunomagnetic sorting (MACS) or fluorescence activated cell sorting (FACS) of CD11b⁺ cells with low or high CD45 expression allowed distinguishing between CD11b⁺CD45^lo (microglia) and CD11b⁺CD45^hi (BM-macrophages) cells from rodent diseased brains [37]. Transcriptomic analyses of immunosorted microglia and macrophages from the rat ischemic brains showed the pro-inflammatory phenotype of microglia over the course of ischemia and the transient influx of the pro-regenerative macrophages into the ischemic brains [38]. With the use of chimeric mice, in which CX3CR1-GFP- monocytes were transplanted to wild-type chimeras, researchers demonstrated distinct contributions of immune cells to the brain maintenance and repair, which led to recognizing distinct roles of microglia and infiltrating BM-derived macrophages [39]. This notion has been supported by the results of conditional ablation of the BM-derived macrophages using the CD11c-DTR system, expressing the human diphtheria toxin receptor (DTR) from the CD11c promoter [40]. The depletion of monocytes/macrophages in CD11b-DTR transgenic mice increased the ischemic lesions and intensified the expression of the inflammatory M1 phenotype markers in CD11b+ cells [41]. The ablation of macrophages in transgenic CD11b-DTR mice had no impact on unilateral traumatic injury lesions, but it led to increases in the expression of pro-inflammatory genes in both hemispheres [42]. Other studies demonstrated that CCR2⁺Ly-6C^hi inflammatory monocytes are rapidly recruited to the CNS of experimental autoimmune encephalomyelitis (EAE) mice, and they are instrumental for the effector phase of disease. The selective depletion of this monocyte subpopulation with an antibody neutralizing CCR2 strongly reduced disease symptoms [43].

Microglia are present at the retina, express the CX3C chemokine receptor 1 (CX3CR1), and undergo morphological activation after corneal injury. Fate-mapping using CX3CR1^+/EGFP::CCR2^+/RFP reporter mice and BM chimeras confirmed that peripheral monocytes/macrophages do not enter into retina under physiological conditions [44]. When busulfan-induced myelodepletion was followed by BM transplantation, peripheral CCR2⁺ CX3CR1⁺ monocytes migrated to the optic nerve but not to the retina under steady-state conditions. Ocular injury led to population of the retina by peripheral CCR2⁺ CX3CR1⁺ monocytes that differentiated to microglia-like CCR2⁻ CX3CR1⁺ cells. Increased monocyte/macrophage trafficking causes microglia activation and elevation of inflammation [44]. After the depletion of microglia with CSF1R inhibitor (PLX5622), even in the absence of ocular injury, peripheral monocytes repopulated the retina. After ocular injury, the engrafted peripheral monocytes were resistant to CSF1R inhibitor and retained a pro-inflammatory phenotype, expressing high levels of MHC-II, interleukin 1 beta (IL-1β), and tumor necrosis factor α(TNF-α) twenty weeks after the injury [45].

All the results point to the heterogeneity of myeloid infiltrates in CNS lesions and the distinct functions of microglia and macrophages [46]. A major challenge in the functional analysis of microglia in diseases is the lack of good experimental systems that allow discriminating between microglia and monocyte-derived macrophages. While the depletion of microglia or macrophages with CD11b-based approaches and other myeloid marker genes provided some interesting clues, these models lack microglial specificity and target other CNS and peripheral cell types. Malignant glioma, a common brain tumor, is another CNS disease, in which a complexity of a tumor microenvironment, infiltrated with numerous immune cells consisting up to 30% of a tumor mass, presents a challenge. Immune cells infiltrating gliomas consist of microglia, BM-derived monocytes, granulocytes, myeloid-derived suppressor cells (MDSCs), and T lymphocytes. The predominant population is glioma-associated microglia and macrophages (GAMs) that accumulate in high-grade gliomas, and their numbers inversely correlate with a patient survival (reviewed in ^[47][48][47,48]). Although GAMs have a few innate immune functions intact, their ability to secrete cytokines and upregulate co-stimulatory molecules is not sufficient to initiate anti-tumor immune responses. Moreover, tumor-reprogrammed GAMs release immunosuppressive cytokines and chemokines blocking anti-tumor responses. GAMs may contribute to tumor progression in different ways: by promoting genetic instability, supporting cancer stem cells, priming invasion, and taming anti-tumor immunity (reviewed in ^[47][48][47,48]). Cell transplantation studies using head-protected irradiation have shown a massive influx of the donor-derived myeloid cells in murine gliomas [49]. Peripheral monocytes/macrophages were detected in gliomas by flow cytometry as Ly6C/MHCII/MerTK/CD64 positive cells [50]. Immunofluorescence studies of platelet-derived growth factor B (PDGFB)-driven gliomas and GL261 gliomas developing in Cx3cr1GFP⁺Ccr2RFP⁺ transgenic mice showed a different location of cells: macrophages (GFP⁺RFP⁺) are predominant in the tumor core, while microglia (GFP⁺RFP^-) accumulate in a tumor periphery [51].

We and others have addressed the issue of heterogeneity and functions of GAMs (usually isolated as CD11b⁺ cells) using bulk transcriptomic approaches, but those studies demonstrated conflicting results [52]. While our studies suggested the pro-tumor phenotype (resembling the wound healing and immunosuppressive M2 phenotype) of GAMs from rodent and human gliomas ^[53][54][55][53,54,55], other studies showed mixed M1/M2-like phenotypes of GAMs from human glioblastomas [56], a mixed inflammatory/anti-tumor phenotype in mouse GL261 gliomas [57], or M0 phenotype in human glioblastomas [58]. Lineage tracing studies provided direct evidence for the contribution of BM-derived macrophages to GAMs in murine gliomas [59]. However, the pressing issue of a cell type and functional diversity of GAMs could not be resolved with classical methods.

The heterogeneity of microglia in various regions, distinctive functions, and contribution of BM-derived monocytes/macrophages and specific roles of these subpopulations in health and disease could not be solved with the traditional methods. Single-cell studies provided a breakthrough and novel insights into the diversity of microglia in health and disease. Mass cytometry or CyTOF (Cytometry by Time-Of-Flight) is a novel platform for high-dimensional phenotypic and functional analysis of single cells. This system uses elemental metal isotopes conjugated to monoclonal antibodies to evaluate over 40 parameters simultaneously on individual cells with minimal overlap between channels [60]. Single-cell RNA sequencing (scRNA-seq) permits determining the entire transcriptome of thousands of individual cells. In recent years, scRNA-seq has been used to study several different tissues and organs, both during development and at a fixed point in time [61] (Table 1).